資源描述:
《人工神經(jīng)網(wǎng)絡(luò)在短期電力負荷預(yù)測中應(yīng)用的研究》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在學術(shù)論文-天天文庫。
1、摘要短期電力負荷預(yù)測是電力系統(tǒng)運行調(diào)度中一項非常重要的內(nèi)容,它是保證電力系統(tǒng)安全經(jīng)濟運行和實現(xiàn)電網(wǎng)科學管理及調(diào)度的重要方面,是能量管理系統(tǒng)(EMS)的組成部分,也是今后進行電網(wǎng)商業(yè)化運營所必需的基本內(nèi)容。本文首先對負荷預(yù)測的現(xiàn)有方法進行了綜述:其次深入研究了神經(jīng)網(wǎng)絡(luò)的模型建立問題,給出了較為適用的建模方法,構(gòu)建了一個三層豹BP神經(jīng)網(wǎng)絡(luò),特別是對BP網(wǎng)絡(luò)模型建立中的隱含層數(shù)確定、隱含層節(jié)點數(shù)確定、訓練次數(shù)與精度的關(guān)系、學習速率的選擇、初始權(quán)值、訓練樣本的選擇及歸一化處理等相關(guān)問題進行了較深入定性和定量分析,并通過算例
2、進行了比較實驗,得出有益結(jié)論。再次在分析BP網(wǎng)絡(luò)缺陷的基礎(chǔ)上,采用改進的BP神經(jīng)網(wǎng)絡(luò)算法,建立了短期負荷預(yù)測的模型,并應(yīng)用改進的BP網(wǎng)絡(luò)算法進行了負荷預(yù)測,比較不同算法的預(yù)測結(jié)果。最后給出了顧及氣象參數(shù)的神經(jīng)網(wǎng)絡(luò)模型建立的方法,通過算例進行了驗證和分析。關(guān)鍵詞:短期負荷預(yù)測;人工神經(jīng)網(wǎng)絡(luò);BP模型改進算法摘要ABSTRACTShort-TermLoadForecasting(STLF)isoneofthemostimportantcontentsofrunninganddispatchingpowersystem.
3、Itisaveryimportantaspectofpowersystemtoensureoperatingsafelyeconomicallyandachievescientificmanagementinthepowersystem.Anditisonepartofenergymanagementsystemaswellasanecessarycontentoftheelectricitymarketplaceoperationmanagement.Thispaperfirstlygivesasummaryfo
4、rpresentmethodofloadforecasting;Secondly,ithasmadeain-depthresearchintoANNmodelingproblem,togivemoreapplicablemodelingmethodandprinciple;Afterstudyingplentyofdocumentsandanalyzingvariousimportantfactorsofelectricpowerload,athree-tierBPneuralnetworkshasbeencons
5、tructed.ToestablishBPnetworkmodel,implicitlayersnumberidentification,hiddennodesdetermination,thetimesandaccuracyoftraining,learningrateoption,theinitialweights,thechoiceoftrainingsamplesandnormalizedtreatment,andotherrelatedissuesareundermorein—depthqualitati
6、veandquantitativeanalysis.Aftermakingcomparedexperimentthroughexamples,usefulconclusionsaredrawn.WiththeanalysisonBPdeficiencies,onthebasisoftheadoptionofimprovedBPneuralnetworkalgorithm,ashort-termloadforecastingmodelisestablished.Theimi)rovedBPnetworkalgorit
7、hmisappliedinloadforecasting.Consequentlythedifferentalgorithms’differentpredictingresultsalecompared.Finally,themethodstoestablishtheneuralnetworkmodeltakenintoaccountofthemeteorologicalparametersarediscussed.Experimentresearchisputforward,thusaverificationan
8、danalysisiSobtained.KeyWords:powersystem,STLF,ANN,advancedalgorithmofBP學位論文獨創(chuàng)性聲明本人聲明所呈交的學位論文是本人在導(dǎo)師指導(dǎo)下進行的研究工作及取得的研究成果。據(jù)我所知,除了文中特別加以標注和致謝的地方外,論文中不包含其他人已經(jīng)發(fā)表或撰寫過的研究成果,也不包含為獲得直昌盍堂或其他教育機